The Expert RecordThe standard for expert witnesses using AI

AI in the Courtroom · Expert witnesses and AI

Watson Grinding explosion trial (3M expert)

An engineering expert retained by 3M in Houston explosion litigation used ChatGPT to draft most of his report and produced about 350 pages of chat logs. He was examined on them at trial. The jury found 3M 30% responsible.

Reported; document not locatedExamined at trial on ChatGPT logs; no ruling on AI use August 2026

What happened

A January 2020 explosion at the Watson Grinding and Manufacturing facility in Houston killed 3 people and damaged about 200 nearby properties. Residents sued several parties, including 3M, over a gas detection system at the site. For one of the trials, 3M retained Josh Autenrieth of KnightHawk Engineering to give expert testimony on the standard of care. This entry rests on news reporting. No court filing was located.

According to 404 Media, which first reported the story, plaintiffs' counsel Will Moye noticed a document titled "Citation Overlay" in discovery and demanded the prompts used to prepare the report. About 350 pages of ChatGPT conversations were produced. The reported prompts include requests to "create an exceptional expert witness report defending the standard of care at 3M" and to "show how 3M is 0% at fault for the explosion." Reports also describe the expert uploading case materials and a photograph of a gas detector and asking the system "what am I looking at?"

What the court did

No ruling excluding or striking the expert's testimony was reported. The expert testified at trial and was examined on the chat logs. Reporting describes plaintiffs' counsel Will Moye calling him in the plaintiffs' case in chief rather than waiting for cross-examination, so the fullest account of the prompts came on direct.

The sequence reported across accounts begins with a denial. Asked at his deposition of July 21, 2026 about his use of the system in preparing the Watson Grinding report, he is quoted as answering "AI did not write my report, that's incorrect," and describing the system's role as "formatting and everything else." As reported, counsel and the expert later agreed on the stand that the report was "90 to 85 percent ChatGPT," and he testified that "AI helped me to draft a straw man to build off of" and pointed to his more than 20 years of industry experience. Asked at trial about the prompts, he is quoted as saying "I don't believe that was wrong to do because I had an opinion already," and, on the wording of the prompts themselves, "The prompts were biased because I had opinions." No tribunal made a finding about any of these answers.

The Harris County jury awarded $61.5 million to 24 plaintiffs and found 3M 30% responsible for the explosion, according to the plaintiffs' firms' announcement of August 11, 2026. That announcement does not mention the expert or AI, and nothing in the sources reviewed ties the verdict to the expert's testimony.

Where it stands

The jury returned a verdict for 24 plaintiffs reported precisely as $61,578,273.19, finding 3M 30% responsible. Sources differ on the date: a Business Wire release gives August 10, 2026, while the plaintiffs' firm and Houston Public Media give August 11, 2026.

The expert is identified in reporting as Josh Autenrieth of Knighthawk Engineering. 3M stated that it disagrees with the verdict and intends to appeal. No motion for new trial, motion notwithstanding the verdict, or notice of appeal was located, and it could not be determined whether judgment has been signed, which is when the thirty-day Texas new-trial deadline begins to run. Another trial was reported as set for October 2026, with roughly 2,000 claims still pending.

This is the third plaintiffs' verdict against 3M in this litigation rather than the third trial. The reported sequence is: a first bellwether settled before trial; a $37.9 million verdict in June 2025; a $118.7 million verdict on November 13, 2025, with 3M found 49% responsible; a defense verdict clearing 3M in a business-loss trial on May 7, 2026; and then this verdict.

The trial is reported as Cause No. 2020-05505-A in the 11th District Court of Harris County, before Judge Kristen Brauchle Hawkins, with Judge Tanya Garrison substituting during final deliberations, under master Cause No. 2021-15294. The caption is reported in one secondary source as Laake v. 3M Co. and has not been confirmed elsewhere. The Harris County District Clerk now requires a registered login for civil case records, so none of this has been confirmed against the docket itself.

Update, October 2026. The ChatGPT material produced in the Watson Grinding litigation has been located. 404 Media published a 365-page production titled AI Prompts (1), Bates-numbered AUTENRIETH 014225 to 014589, on DocumentCloud; it includes the prompt asking the system to "create an exceptional expert witness report defending the standard of care of 3M" and to show that 3M was 0% at fault. This is a discovery production, not a court filing. A plaintiffs' firm release gives the consolidated proceeding as Cause No. 2020-05505-A. No judgment, post-trial motion or notice of appeal was located as of October 8, 2026.

The obligation it engaged

The reported prompts engage Clause 1, sequence, and Clause 2, non-delegation. A prompt asking a system to show that a party is 0% at fault states the conclusion before the analysis, and a report that is mostly generated text leaves the court to ask which opinions are the expert's. The expert's answer at trial, that the system drafted a straw man he built on, is a claim that Clause 2 was met. The logs were the evidence used to test that claim.

The matter also shows Clause 7, contemporaneous record, operating in an unexpected direction. The chat history served as a complete record of the process, and once produced it became the main subject of cross-examination. An expert whose use of a system is ordered correctly under Clause 1 has little to fear from producing such a record. Clause 8, disclosure, applies because the use came to light through discovery rather than from the report itself.

Sources

What could not be confirmed

News only; no court filing located. The 404 Media article is paywalled and only its opening was read; trial and deposition quotations come from outlets summarizing it. Accounts differ on how the logs were obtained: the better corroborated account is a demand at deposition followed by a three-hour pause while the logs were gathered, while Futurism alone describes the judge ordering disclosure. Reported figures differ: the share of the 30-page report attributed to the system is variously 'more than 80 percent,' '90 to 85 percent' and 85 to 90 percent, and the volume of logs produced is given as 300, 350 and 365 pages. The award is reported precisely as $61,578,273.19 by the plaintiffs' firms and as $62.3 million by The Texas Lawbook. Sources differ on the verdict date, August 10 or 11, 2026. The cause number, caption, court and judge above come from reporting and have not been confirmed against the Harris County District Clerk's records, which now require a registered login; the Laake v. 3M Co. caption rests on a single secondary source. Two cause numbers circulate in this litigation and both are correct rather than one being an error: 2020-05505-A is this August 2026 trial, while 2020-05585-B is Inez Vasquez-Lopez v. 3M Company, the November 2025 trial in the 295th Judicial District Court before Judge Donna Roth. CBS News gives the expert's first name as Jason; the name used here, Josh, is the one used by KnightHawk Engineering's own hiring announcement and by every other outlet located.

About this entry

The primary document had not been located when this entry was checked, so it rests on the reporting listed above. The entry for Watson Grinding explosion trial records conduct and what the tribunal did. It does not characterize a holding and it is not legal advice. Last checked October 8, 2026. More matters like this one are collected under the expert witness matters. If something here is out of date, write to [email protected].

Top